Included in this repo is a project that utilizes clustering and constrained optimization to create an ideal roster for the Detroit Red Wings. First, k-means clustering is used to cluster NHL players for the 20-21, 21-22, and 22-23 seasons based on playing style. Next, Detroit's 22-23 roster was evaluated based on clustering results. Finally, constrained optimization was used to identify the optimal roster considering player cluster types, positions, trades, and salary cap. From the optimization results, recommendations were made as to who the Wings should keep on their roster for the next season and who they should attempt to trade away.
Please be sure to check out the write-up in 'documentation' for a detailed report of data acquisition, methods, results, analyses, and recommendations!
Data for this project was acquired from capfriendly.com and hockey-reference.com.
Note that due to capfriendly.com's continual updates of the free agent page, a user will not receive identical results to those in the report.